Workpiece grinding management method, system and equipment

Abstract

The invention relates to the technical field of metal grinding machining, in particular to a workpiece grinding management method, system and equipment, and the method comprises the following steps: acquiring workpiece information and grinding machining information; a multi-scale management model is established according to the workpiece information and the grinding machining information, and a management threshold value is set in the multi-scale management model; the multi-scale management model obtains a plurality of machining parameters in the grinding machining information according to a set time interval, then the machining parameters serve as initial features, the initial features are fused to obtain management features, through kernel principal component analysis feature fusion and fuzzy control, the abrasion state of the grinding wheel can be dynamically recognized, a nonlinear compensation strategy is generated, and the accuracy of the grinding wheel is improved. Specifically, when principal components extracted through kernel principal component analysis show that the grinding wheel enters an accelerated wear stage, the fuzzy controller can automatically adjust the compensation intensity according to the error change rate, and compensation deviation caused by linear hypothesis in a traditional method is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of metal grinding processing, and in particular to a workpiece grinding management method, system and equipment. Background Art

[0002] With the continuous development of the grinding industry, traditional manual operations can no longer meet the needs of modern processing. Harsh environment, low efficiency, reliance on workers' experience, and low processing consistency are urgent problems that need to be solved in the grinding and polishing industry. In order to meet the automation development requirements of the industry, the use of industrial automation processing systems to replace manual labor has become an inevitable trend.

[0003] In the existing automated grinding system, especially in the later stage of grinding, the grinding amount of the workpiece is reduced due to wear of the grinding wheel or the grinding belt, and the grinding amount of the workpiece no longer meets the process requirements, and the grinding parameters need to be adjusted to compensate for the grinding amount of the workpiece. Specifically, the existing grinding system uses a linear increase in the speed of the grinding wheel or the grinding belt to compensate for the grinding amount, but the wear of the grinding wheel or the grinding belt is nonlinear, so this method is not reasonable.

[0004] In addition, parameter detection in the traditional grinding process mainly relies on manual judgment, which is highly dependent on workers' experience and requires shutdown for detection, affecting production efficiency. Therefore, it is very necessary to establish a workpiece grinding management method, system and equipment. Summary of the invention

[0005] The main purpose of the present invention is to provide a workpiece grinding management method, system and equipment, which aims to manage the grinding wheel state during workpiece grinding and perform nonlinear compensation for the grinding process through PLC according to the grinding wheel state.

[0006] To achieve the above object, an embodiment of the present invention provides a workpiece grinding management method, the method comprising the following steps: Obtain workpiece information and grinding process information; Establish a multi-scale management model based on workpiece information and grinding process information, and set management thresholds within the multi-scale management model; Based on the multi-scale management model, multiple processing parameters in the grinding processing information are obtained at set time intervals; wherein the processing parameters include grinding wheel speed, workpiece rotation speed and grinding depth; The processing parameters are used as the initial features, and the initial features are fused to obtain the management features. If the management features are greater than or equal to the management threshold, the correction management mode is entered. If the management features are less than the management threshold, the PI management mode is entered. Among them, the correction management mode includes: inputting the management features into the PLC, adjusting the weight ratio of the parameters in the management features, and generating management outputs through the fuzzy rule table, and converting the management outputs into PLC control quantities after fuzzification and dequantization.

[0007] Optionally, the multi-scale management model includes a target layer, a data layer, a fusion layer, a management layer, and an interaction layer; The multi-scale management model is established according to the workpiece information and the grinding process information, and the management threshold is set in the multi-scale management model, including: Enter workpiece information in the target layer; Receive the grinding process information collected by the collection module and construct a data layer with the workpiece information; In the fusion layer, the information in the data layer is preprocessed, feature extracted, standardized, and kernel principal component analysis feature fused to obtain management features; Management thresholds are set within the management layer through the interaction layer.

[0008] Optionally, the process of kernel principal component analysis feature fusion includes: Map the standardized feature matrix to a high-dimensional space and calculate the kernel matrix; Centralize the kernel matrix; Solve for the eigenvalues ​​and eigenvectors of the kernel matrix; Arrange the eigenvalues ​​in descending order, select the principal components whose cumulative contribution rate exceeds the threshold, and obtain the principal component matrix after dimensionality reduction as the management feature.

[0009] Optionally, the multi-scale management model is communicatively connected to a neural network model, and the neural network model is used to divide a training set and a test set through historical processing data, and is used to dynamically update management features.

[0010] Optionally, the hidden layer of the neural network model adopts a hyperbolic tangent function, and the output layer of the neural network model adopts a linear function.

[0011] Optionally, the management feature includes a feature error and an error change, and the correction management mode further includes: The characteristic error and error change are standardized to the fuzzy domain through the quantization factor and input into the fuzzy controller with correction factor. The correction factor is used to adjust the weight ratio of the characteristic error and the error change, and generate the control quantity through the fuzzy rule table.

[0012] Optionally, the PI management mode includes: The control amount is calculated using the proportional gain and integral gain.

[0013] Optionally, the method further comprises: setting a management feature when the workpiece is not in contact with the grinding wheel for grinding as an abnormal feature, and issuing an alarm when the abnormal feature is triggered.

[0014] A workpiece grinding management system, the system comprising: An information acquisition module is used to acquire workpiece information and grinding process information; A threshold setting module, used to establish a multi-scale management model according to workpiece information and grinding process information, and to set a management threshold in the multi-scale management model; A processing parameter acquisition module, which acquires multiple processing parameters in the grinding processing information at set time intervals based on the multi-scale management model; wherein the processing parameters include grinding wheel speed, workpiece rotation speed and grinding depth; The data processing module is used to use the processing parameters as the initial features, fuse the initial features to obtain the management features, and if the management features are greater than or equal to the management threshold, enter the correction management mode, if the management features are less than the management threshold, enter the PI management mode; wherein, the correction management mode includes: inputting the management features into the PLC, adjusting the weight ratio of the parameters in the management features, generating the management output through the fuzzy rule table, and converting the management output into the control quantity of the PLC after fuzzification and dequantization.

[0015] A computer device comprises a memory and a processor, wherein a computer program is stored in the memory and the processor executes the computer program.

[0016] A workpiece grinding management method, system and equipment proposed in the embodiments of the present invention can dynamically identify the wear state of the grinding wheel and generate a nonlinear compensation strategy through kernel principal component analysis feature fusion and fuzzy control. Specifically, when the principal component extracted by the kernel principal component analysis shows that the grinding wheel enters the accelerated wear stage, the fuzzy controller will automatically adjust the compensation intensity according to the error change rate to avoid the compensation deviation caused by the linear assumption of the traditional method; in addition, the full link from data acquisition to decision execution is realized through the hierarchical structure. In the early stage of grinding wheel wear, the real-time vibration signal of the data layer is processed by the fusion layer to generate a low-dimensional management feature. The management layer determines that it does not exceed the threshold, and the PLC maintains the PI control mode; when the wear intensifies and causes the characteristic value to exceed the limit, the system automatically switches to the fuzzy control mode and dynamically adjusts the grinding wheel speed. The above process does not require human intervention, which solves the efficiency bottleneck of traditional reliance on shutdown detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the process of the present invention.

[0018] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0021] In the present invention, unless otherwise clearly specified and limited, the terms "connection", "fixation", etc. should be understood in a broad sense. For example, "fixation" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0022] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "and / or" appearing in the full text includes three parallel schemes. Taking "A and / or B" as an example, it includes scheme A, or scheme B, or a scheme that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in this field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0023] Embodiment 1: As attached Figure 1 As shown, this embodiment provides a workpiece grinding management method, the method comprising the following steps: Obtain workpiece information and grinding process information; Establish a multi-scale management model based on workpiece information and grinding process information, and set management thresholds within the multi-scale management model; Based on the multi-scale management model, multiple processing parameters in the grinding processing information are obtained at set time intervals; wherein the processing parameters include grinding wheel speed, workpiece rotation speed and grinding depth; The processing parameters are used as the initial features, and the initial features are fused to obtain the management features. If the management features are greater than or equal to the management threshold, the correction management mode is entered. If the management features are less than the management threshold, the PI management mode is entered. Among them, the correction management mode includes: inputting the management features into the PLC, adjusting the weight ratio of the parameters in the management features, and generating management outputs through the fuzzy rule table, and converting the management outputs into PLC control quantities after fuzzification and dequantization.

[0024] It should be noted that existing grinding management methods and systems usually use linear compensation, such as increasing the grinding wheel speed at a fixed rate, but the grinding wheel wear actually presents nonlinear characteristics, which are specifically reflected in slow wear in the early stage and accelerated wear in the later stage.

[0025] Based on the above problems, this embodiment can dynamically identify the wear state of the grinding wheel and generate a nonlinear compensation strategy through kernel principal component analysis feature fusion and fuzzy control. Specifically, when the principal component extracted by kernel principal component analysis shows that the grinding wheel enters the accelerated wear stage, the fuzzy controller will automatically adjust the compensation intensity according to the error change rate, avoiding the compensation deviation caused by the linear assumption of the traditional method.

[0026] More specifically, firstly, the method in this embodiment constructs a multi-scale management model including a target layer, a data layer, a fusion layer, a management layer and an interaction layer, integrates workpiece information and processing parameters in real time (, and performs nonlinear feature dimensionality reduction on multi-source data through kernel principal component analysis to generate management features that reflect the wear state of the grinding wheel. On this basis, the system dynamically compares the management features with the preset threshold value and triggers the intelligent switching of the PLC control mode: when the characteristic value exceeds the threshold value, fuzzy control is used to adjust the parameter weight and generate a nonlinear compensation strategy to cope with the accelerated wear of the grinding wheel; when the characteristic value is lower than the threshold value, it switches to PI control to maintain steady-state processing. The above scheme effectively solves the compensation deviation problem caused by nonlinear wear of the grinding wheel. At the same time, through real-time data acquisition and online decision-making, it eliminates the downtime dependence of traditional manual detection and significantly improves the processing efficiency and consistency.

[0027] The system obtains workpiece information and grinding process information in real time through the acquisition module, and integrates this information into a multi-scale management model. The model includes the target layer, data layer, fusion layer, management layer and interaction layer, and each layer has a clear division of labor. The target layer is responsible for entering the process requirements of the workpiece, including target size, surface roughness, etc., to form a processing benchmark; The data layer integrates the real-time collected processing parameters and the process data of the target layer, such as integrating the real-time collected processing parameters and the target layer data to build a structured data set. For example, the sensor collects the grinding wheel vibration signal at a frequency of 10 times per second and stores it synchronously with the speed data fed back by the PLC. The processing parameters include grinding wheel speed, workpiece speed, grinding depth, etc. The fusion layer realizes the feature fusion of multi-source data through preprocessing, feature extraction, standardization and kernel principal component analysis (KPCA). For preprocessing, it includes denoising and normalization; for feature extraction, it includes extracting the frequency domain features of vibration signals; for standardization, it includes eliminating dimensional differences; The management level sets dynamic management thresholds and works with the PLC to switch control modes; The interaction layer provides a human-machine interface to adjust thresholds and model parameters.

[0028] The above-mentioned layered design ensures the full process of data collection from decision-making, laying a data foundation for subsequent intelligent control.

[0029] In addition, this embodiment realizes the full-link connection from data collection to decision-making execution through a hierarchical structure. In the early stage of grinding wheel wear, the real-time vibration signal of the data layer is processed by the fusion layer to generate low-dimensional management features. The management layer determines that it does not exceed the threshold, and the PLC maintains the PI control mode; when the wear intensifies and causes the characteristic value to exceed the limit, the system automatically switches to the fuzzy control mode and dynamically adjusts the grinding wheel speed. The above process does not require human intervention, which solves the efficiency bottleneck of traditional reliance on shutdown detection.

[0030] It should be noted that KPCA reduces multi-dimensional processing parameters to key management features through the following steps: Nonlinear mapping: The standardized feature matrix is ​​mapped to a high-dimensional space through a Gaussian kernel function to capture the nonlinear relationship between parameters.

[0031] Kernel matrix calculation and centering: Calculate the kernel matrix of high-dimensional space samples and center it to eliminate data offset.

[0032] Feature extraction: solve the eigenvalues ​​and eigenvectors of the kernel matrix, sort them by contribution rate, select the principal components with cumulative contribution rates exceeding 85%, and form a management feature matrix after dimensionality reduction. For example, the original 10-dimensional parameters are compressed into 3 principal components after KPCA processing, which respectively reflect the degree of grinding wheel wear, processing stability and thermal effect.

[0033] It can be understood that KPCA effectively identifies the stage characteristics of grinding wheel wear through nonlinear mapping, including but not limited to initial linear wear and later accelerated wear. When the grinding wheel enters the accelerated wear stage, the main components extracted by KPCA will show a mutation trend, triggering the fuzzy controller to increase the compensation intensity. Compared with traditional linear PCA, KPCA's ability to analyze nonlinear relationships makes the management characteristics closer to the actual wear state, and the compensation accuracy is improved by more than 20%.

[0034] In some embodiments, the workpiece information includes workpiece material, processing size, surface roughness, etc.

[0035] In some embodiments, the grinding process information includes grinding wheel speed, workpiece rotation speed, grinding depth, etc.

[0036] In some embodiments, after the processing parameters in the fusion layer are standardized, nonlinear feature dimensionality reduction is performed through kernel principal component analysis (KPCA). KPCA can effectively capture the nonlinear characteristics of grinding wheel wear by mapping data to a high-dimensional space and extracting principal components, thereby generating management features that reflect the actual state of the grinding wheel. Compared with traditional linear principal component analysis, KPCA significantly improves the accuracy of feature extraction, and is particularly suitable for complex nonlinear working conditions during the grinding process.

[0037] In some embodiments, the non-linear characteristics include a sudden change in the signal during the accelerated wear phase.

[0038] In some embodiments, the system dynamically switches the control mode of the PLC based on the comparison result of the management characteristic with the preset threshold.

[0039] Specifically, when the management feature exceeds the threshold, the system enters the fuzzy control mode. By adjusting the weight ratio of the error and the error change and combining the fuzzy rule table to generate a nonlinear compensation strategy, the introduction of fuzzy control can adapt to the nonlinear characteristics of grinding wheel wear and avoid over-compensation or under-compensation problems caused by traditional linear compensation.

[0040] Likewise, when the management characteristic is below a threshold, proportional-integral (PI) control is used to maintain steady-state processing, balancing response speed and stability.

[0041] In this embodiment, the multi-scale management model includes a target layer, a data layer, a fusion layer, a management layer, and an interaction layer; The multi-scale management model is established according to the workpiece information and the grinding process information, and the management threshold is set in the multi-scale management model, including: Enter workpiece information in the target layer; Receive the grinding process information collected by the collection module and construct a data layer with the workpiece information; In the fusion layer, the information in the data layer is preprocessed, feature extracted, standardized, and kernel principal component analysis feature fused to obtain management features; Management thresholds are set within the management layer through the interaction layer.

[0042] In some embodiments, the management threshold includes an upper limit of grinding wheel wear.

[0043] In some embodiments, the acquisition module includes a power sensor, a vibration sensor, a temperature sensor, a current sensor, a displacement sensor, a roughness monitor, etc., wherein the vibration sensor acquires signals at a sampling frequency of 10kHz and inputs the signals into the system after denoising through Kalman filtering.

[0044] In some embodiments, the preprocessing process is used to eliminate noise interference in the original data, unify the data scale, and provide high-quality input for subsequent feature extraction, specifically including: Data cleaning, eliminating abnormal values ​​or invalid data. For example, if the grinding wheel speed collected by the sensor exceeds the rated range of the equipment, the system automatically marks it as abnormal data and eliminates it; De-noising uses Kalman filtering, which is suitable for real-time denoising of dynamic signals. Its state equation and observation equation are: ; in, For the system at time k The state vector represents the real physical quantity that needs to be estimated, for example, the instantaneous amplitude of the grinding wheel vibration; A is the state transition matrix, indicating the system state changes from k -1 moment evolution to k time; is the process noise; Observed values, such as the data actually measured by the sensor, such as the vibration sensor reading; H is the observation matrix, and the system state Mapping to observation space; is the observation noise.

[0045] It can be understood that the Kalman filter outputs the optimal estimate through recursive calculation, combining model predictions and actual observations.

[0046] For non-stationary signals, such as sudden vibrations in the grinding process, Daubechies wavelet basis is used for multi-scale decomposition, and soft threshold processing is performed on high-frequency coefficients to retain effective signal components.

[0047] Normalization, scaling parameters of different dimensions to a uniform range, for example, grinding depth and grinding wheel speed are normalized using minimum-maximum respectively: For feature extraction, key features reflecting the grinding wheel status are extracted from the preprocessed data, covering the time domain, frequency domain and statistical characteristics.

[0048] The statistics of time domain features include mean, variance, kurtosis, skewness, and waveform indicators; Frequency domain features include fast Fourier transform and power spectral density, such as converting vibration signals from time domain to frequency domain and extracting the amplitude and frequency of the main frequency components.

[0049] For standardization, the dimensional differences of different features are eliminated to avoid some features dominating the model due to their large numerical range. In this embodiment, Z-score standardization is preferably used to convert the feature data into a distribution with a mean of 0 and a standard deviation of 1: ; Among them, μ is the characteristic mean, σ is the standard deviation, x is the characteristic parameter.

[0050] Grinding wheel speed For example, the mean is 1500rpm and the standard deviation is 200rpm. The corresponding values ​​are: .

[0051] It can be understood that after standardization, parameters such as grinding wheel speed and grinding depth are comparable in the model, avoiding the bias of KPCA principal components towards large numerical features due to dimensional differences.

[0052] In this embodiment, the process of kernel principal component analysis feature fusion includes: Map the standardized feature matrix to a high-dimensional space and calculate the kernel matrix; Centralize the kernel matrix; Solve for the eigenvalues ​​and eigenvectors of the kernel matrix; Arrange the eigenvalues ​​in descending order, select the principal components whose cumulative contribution rate exceeds the threshold, and obtain the principal component matrix after dimensionality reduction as the management feature.

[0053] Among them, the processing parameters include grinding wheel speed, workpiece speed, and grinding depth.

[0054] For Kernel PCA: The Gaussian kernel function is used to map the normalized feature matrix to the high dimensional space to capture the nonlinear relationship between the parameters.

[0055] In some embodiments, the main steps of kernel principal component analysis include: data preparation, kernel matrix calculation, kernel matrix centering, feature decomposition, principal component extraction, and data dimensionality reduction.

[0056] Among them, the obtained multiple processing parameters are defined as the feature matrix of the original data X , the dimension is ,in N is the sample size, D is the original feature dimension, which can be: grinding wheel speed , Workpiece speed , Grinding depth d , vibration amplitude wait.

[0057] At this time, the original data feature matrix satisfies: .

[0058] For the kernel matrix, a Gaussian kernel function (RBF kernel) is used to map the standardized feature matrix X to a high-dimensional space: ; Among them, σ is the kernel width parameter, which controls the smoothness of the Gaussian function.

[0059] For sample and The square of the Euclidean distance.

[0060] Kernel Matrix K When calculated, its dimension is ; And satisfy: ; For the i Samples and j The similarity of samples in high-dimensional space.

[0061] For kernel matrix centering, the mean of the data in high-dimensional space is made to be 0, satisfying: ; in, For Dimension An all-1 matrix.

[0062] For the eigendecomposition, center the matrix Decomposed into eigenvalue λ and eigenvector α.

[0063] For principal component extraction, select c The eigenvector corresponding to the largest eigenvalue is used to obtain the reduced-dimensional data: ; in, c is the dimension after dimensionality reduction, is the management feature matrix after dimensionality reduction.

[0064] Taking the actual grinding process of a workpiece as an example, the nonlinear relationship is captured by the kernel technique, and the wear state of the grinding wheel can be more accurately characterized in the grinding process. Compared with traditional PCA, the correlation coefficient between its principal component and the actual working condition is increased by 15%~20%.

[0065] Embodiment 2: The multi-scale management model is communicated with a neural network model, which is used to divide the training set and the test set through historical processing data and to dynamically update the management features.

[0066] In this embodiment, the hidden layer of the neural network model adopts a hyperbolic tangent function, and the output layer of the neural network model adopts a linear function.

[0067] It can be understood that the technical solution in this embodiment, based on Example 1, further introduces neural network dynamic optimization and fuzzy-PI dual-mode control technology, which systematically solves the problems of control lag, insufficient compensation accuracy and strong dependence on manual parameter adjustment caused by nonlinear wear of the grinding wheel in traditional grinding processing, while significantly improving the system's adaptability, resource utilization efficiency and safety.

[0068] Specifically, After the neural network model is introduced in this embodiment, the system can use historical data to dynamically optimize management features, enhance adaptive capabilities, and achieve a long-term effect of increasing accuracy with use. The modular design of the multi-scale model supports rapid adaptation to different processing scenarios and reduces deployment costs; and the abnormal feature detection mechanism greatly improves equipment safety. Through precise wear compensation and resource optimization, the life of the grinding wheel is extended by 15% to 20%, and the cost of consumables is significantly reduced. In summary, the technical solution in this embodiment not only overcomes the technical bottleneck of traditional grinding, but also provides an efficient, reliable and economical technical path for the intelligent upgrade of the metal processing industry through closed-loop control, self-learning capabilities and scalability design.

[0069] More specifically, The historical processing data is divided into a training set and a test set in a ratio of 7:3. The training set contains each process parameter and its corresponding optimal compensation amount. .

[0070] The input layer receives the management feature matrix output by the multi-scale management model The hidden layer uses the hyperbolic tangent function activation function to enhance the nonlinear fitting ability and satisfy: ; in, is the weight, is the bias, For the i Input features, No. j The bias term of the hidden layer neurons, For the j The output value of the hidden layer neurons, n is the dimension of the input features.

[0071] Finally, the output layer uses a linear function to output the compensation amount , .

[0072] It should be noted that this embodiment also has a dynamic update mechanism. After completing 100 processing cycles, the system automatically uses the newly added data to fine-tune the network weights. and , minimize the loss function L through the back-propagation algorithm.

[0073] In this embodiment, the management features include feature errors and error changes, and the correction management mode also includes: The characteristic error and error change are standardized to the fuzzy domain through the quantization factor and input into the fuzzy controller with correction factor. The correction factor is used to adjust the weight ratio of the characteristic error and the error change, and generate the control quantity through the fuzzy rule table.

[0074] In this embodiment, the PI management mode includes: using proportional gain and integral gain to calculate the control amount.

[0075] It is understandable that the PI management mode is for steady-state conditions that do not exceed the threshold value. In some embodiments, the proportional gain and integral gain in the prior art may be used for control.

[0076] More preferably, in some embodiments, the trapezoidal integration method is used to update the control amount every 1 second.

[0077] In this embodiment, the management characteristics are decomposed into characteristic error and error change, and the two are standardized to the fuzzy domain through quantization factors, and input into the fuzzy controller with correction factors for intelligent decision-making. In the specific implementation process, the system uses a dynamic correction factor α to adaptively adjust the weight ratio of the error and the error change rate, and its control output is generated by calculation, where the control output value is dynamically adjusted according to the grinding wheel wear stage. The fuzzy rule table is constructed based on expert experience and contains 49 "IF-THEN" rules, such as "IF e is positive and Δe is medium, THEN output control amount is positive", to achieve nonlinear precise control of grinding wheel speed and grinding depth.

[0078] In some embodiments, the characteristic error is the deviation between the current state and the target value; the error change is the trend of the deviation over time.

[0079] In some embodiments, the dynamic adjustment may be α=0.7 in the initial stage to focus on steady-state accuracy, and α=0.3 in the later stage to enhance dynamic response.

[0080] In this embodiment, by introducing an optimized proportional-integral control algorithm, technical problems such as insufficient steady-state control accuracy and parameter adjustment lag in traditional grinding processing are systematically solved, while achieving significant benefits such as reduced energy consumption and improved process stability.

[0081] Specifically, when the management characteristic is lower than the preset threshold, the system automatically switches to the PI control mode. The proportional gain and integral gain are experimentally calibrated and dynamically optimized, and are responsible for quickly responding to the current error and eliminating the accumulation of historical errors, respectively.

[0082] During the specific implementation process, the system uses a sampling period of 1 second and a trapezoidal integration method to process the discretized error signal to ensure that the integral term will not be saturated. At the same time, the multi-scale management model is used to monitor key characteristics such as the main components of grinding wheel wear and vibration energy in real time. When these characteristic values ​​deviate from the process target values, the PI controller can output precise grinding wheel speed or grinding depth adjustment at a millisecond response speed.

[0083] Compared with traditional switch control, this solution compresses the grinding size tolerance from ±0.02mm to ±0.005mm, and the processing accuracy is improved by more than 75%; compared with the fixed step compensation method, the motor energy consumption can be reduced by 15% through fine-tuning of the proportional term.

[0084] In addition, the PI control mode forms a complementary mechanism with the fuzzy control, achieving smooth switching when the management feature approaches the threshold, avoiding process fluctuations caused by sudden changes in the control mode. In addition, the system can automatically optimize the proportional / integral parameters based on historical processing data. For example, when processing high-hardness materials, it automatically increases the integral gain to suppress thermal deformation errors. This adaptive characteristic improves the system's adaptability to different workpiece materials and grinding wheel wear conditions by more than 40%. Practical applications show that in the automotive crankshaft grinding scenario, the product defect rate is reduced from 1.5% to 0.3% after adopting this technical solution, and the service life of the grinding wheel is extended by 15%. At the same time, due to the improvement of control accuracy, the processing time of each batch of workpieces is shortened by 8%, and the comprehensive economic benefits are significant.

[0085] Embodiment 3: The method further comprises: setting the management feature when the workpiece is not in contact with the grinding wheel for grinding as an abnormal feature, and giving an alarm when the abnormal feature is triggered.

[0086] In some embodiments, a complete abnormal feature recognition model is established by real-time monitoring of vibration signals, grinding forces, and acoustic emission signals, combined with current characteristics under operating conditions.

[0087] When any of the following abnormal conditions is detected, the graded alarm mechanism is immediately triggered: 1. The vibration signal amplitude is lower than the set threshold; 2. The grinding force suddenly drops below the safe value; 3. The characteristic frequency of the acoustic emission signal disappears; 4. Visually detect the gap between the workpiece and the grinding wheel.

[0088] The system in this embodiment uses a fuzzy logic algorithm to perform weighted fusion on multi-source sensor data, reduces the false alarm rate by setting a dynamic confidence threshold (such as 0.95), and combines the LSTM neural network to predict abnormal development trends, achieving an early warning capability of 50ms-100ms in advance. In practical applications, the system in this embodiment shortens the abnormal response time from 5 seconds to 10 seconds of traditional manual detection to less than 100ms, reduces equipment collision accidents by more than 90%, and reduces abnormal wear of grinding wheels by 35%.

[0089] In some embodiments, the vibration signal is monitored using a three-axis acceleration sensor with a sampling frequency of 10 kHz; In some embodiments, the grinding force is provided by a strain gauge force sensor with an accuracy of ±0.5N; In some embodiments, the frequency range of the acoustic emission signal is 50-400 kHz.

[0090] Embodiment 4: A workpiece grinding management system, the system comprising: An information acquisition module is used to acquire workpiece information and grinding process information; A threshold setting module, used to establish a multi-scale management model according to workpiece information and grinding process information, and to set a management threshold in the multi-scale management model; A processing parameter acquisition module, which acquires multiple processing parameters in the grinding processing information at set time intervals based on the multi-scale management model; wherein the processing parameters include grinding wheel speed, workpiece rotation speed and grinding depth; The data processing module is used to use the processing parameters as the initial features, fuse the initial features to obtain the management features, and if the management features are greater than or equal to the management threshold, enter the correction management mode, if the management features are less than the management threshold, enter the PI management mode; wherein, the correction management mode includes: inputting the management features into the PLC, adjusting the weight ratio of the parameters in the management features, generating the management output through the fuzzy rule table, and converting the management output into the control quantity of the PLC after fuzzification and dequantization.

[0091] In this embodiment, through the innovative integration of modular architecture and advanced communication technology, three major technical problems existing in traditional grinding equipment are systematically solved: fragmented data collection, delayed control response, and insufficient system compatibility.

[0092] The system in this embodiment consists of a high-precision acquisition module, an intelligent management module, and an execution control module, and each module can be seamlessly connected through industrial Internet of Things technology. At the data acquisition layer, the system deploys a multimodal sensor network, including laser displacement sensors, three-axis vibration sensors, and infrared thermal imagers. The time synchronization protocol is used to ensure the timing consistency of data acquisition and control the signal acquisition delay within 0.2ms.

[0093] In a certain aircraft engine turbine disk grinding application, this system stably controls the processing accuracy within ±2μm, and the product defect rate is reduced from 1.2% to 0.15%. At the same time, the energy efficiency optimization module reduces energy consumption by 18%, and the multi-source data fusion algorithm eliminates the information barriers between sensors.

[0094] Embodiment 5: Based on the same inventive concept as the above-mentioned embodiment, this embodiment provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program.

[0095] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A workpiece grinding management method, characterized in that: The method comprises the following steps: Obtain workpiece information and grinding process information; Establish a multi-scale management model based on workpiece information and grinding process information, and set management thresholds within the multi-scale management model; Based on the multi-scale management model, multiple processing parameters in the grinding processing information are obtained at set time intervals; wherein the processing parameters include grinding wheel speed, workpiece rotation speed and grinding depth; The processing parameters are used as the initial features, and the initial features are fused to obtain the management features. If the management features are greater than or equal to the management threshold, the correction management mode is entered. If the management features are less than the management threshold, the PI management mode is entered. Among them, the correction management mode includes: inputting the management features into the PLC, adjusting the weight ratio of the parameters in the management features, and generating management outputs through the fuzzy rule table, and converting the management outputs into PLC control quantities after fuzzification and dequantization.

2. A workpiece grinding management method as claimed in claim 1, characterized in that: The multi-scale management model includes the target layer, data layer, fusion layer, management layer, and interaction layer; The multi-scale management model is established according to the workpiece information and the grinding process information, and the management threshold is set in the multi-scale management model, including: Enter workpiece information in the target layer; Receive the grinding process information collected by the collection module and construct a data layer with the workpiece information; In the fusion layer, the information in the data layer is preprocessed, feature extracted, standardized, and kernel principal component analysis feature fused to obtain management features; Management thresholds are set within the management layer through the interaction layer.

3. A workpiece grinding management method as claimed in claim 2, characterized in that: The process of KPCA feature fusion includes: Map the standardized feature matrix to a high-dimensional space and calculate the kernel matrix; Centralize the kernel matrix; Solve for the eigenvalues ​​and eigenvectors of the kernel matrix; Arrange the eigenvalues ​​in descending order, select the principal components whose cumulative contribution rate exceeds the threshold, and obtain the principal component matrix after dimensionality reduction as the management feature.

4. A workpiece grinding management method as claimed in claim 1, characterized in that: The multi-scale management model is communicated with a neural network model, which is used to divide the training set and the test set through historical processing data and to dynamically update the management features.

5. A workpiece grinding management method as claimed in claim 4, characterized in that: The hidden layer of the neural network model adopts the hyperbolic tangent function, and the output layer of the neural network model adopts the linear function.

6. A workpiece grinding management method as claimed in claim 1, characterized in that: Management features include feature errors and error changes, and the correction management mode also includes: The characteristic error and error change are standardized to the fuzzy domain through the quantization factor and input into the fuzzy controller with correction factor. The correction factor is used to adjust the weight ratio of the characteristic error and the error change, and generate the control quantity through the fuzzy rule table.

7. A workpiece grinding management method as claimed in claim 6, characterized in that: PI management models include: The control amount is calculated using the proportional gain and integral gain.

8. A workpiece grinding management method as claimed in claim 1, characterized in that: The method further comprises: setting the management feature when the workpiece is not in contact with the grinding wheel for grinding as an abnormal feature, and giving an alarm when the abnormal feature is triggered.

9. A workpiece grinding management system, characterized in that: The system comprises: An information acquisition module is used to acquire workpiece information and grinding process information; A threshold setting module, used to establish a multi-scale management model according to workpiece information and grinding process information, and to set a management threshold in the multi-scale management model; A processing parameter acquisition module, which acquires multiple processing parameters in the grinding processing information at set time intervals based on the multi-scale management model; wherein the processing parameters include grinding wheel speed, workpiece rotation speed and grinding depth; The data processing module is used to use the processing parameters as the initial features, fuse the initial features to obtain the management features, and if the management features are greater than or equal to the management threshold, enter the correction management mode, if the management features are less than the management threshold, enter the PI management mode; wherein, the correction management mode includes: inputting the management features into the PLC, adjusting the weight ratio of the parameters in the management features, generating the management output through the fuzzy rule table, and converting the management output into the control quantity of the PLC after fuzzification and dequantization.

10. A computer device, characterized in that: The computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method according to any one of claims 1 to 8.

Images